The clinical trials industry is just coming to grips with a new FDA approach to inspections that uses advanced computer algorithms to flag trials for risk.

The agency has spent the past few years preparing for centralized trial inspections and issued a draft guidance on them in February but it’s still relatively early in the process, says Cerdi Beltre, senior vice president for institutional services at the WIRB-Copernicus Group. Until trial professionals get a better sense of how the algorithm works, they should do some self-auditing, she adds.

Some boxes to check, she says: “How are they ensuring all the sites are compliant? What resources are available? What high-level checklist can they have to prepare for that?”

Beltre’s advice comes on the heels of a presentation by Jean Mulinde, senior policy advisor at CDER’s clinical compliance evaluation division, laying out how the FDA’s “decision tree” algorithm helps the agency determine which trials to inspect.

Mulinde has been briefing sites and sponsors every chance she gets in recent months — including at last week’s Metrics Championship Consortium conference — to help them prepare. But Beltre said it was still eye-opening to hear a top regulator acknowledge that inspections are driven in part by a computer code.

The process is complex — and the FDA is keeping some of its inputs close to the vest — but the essence of it is that the agency is assigning weighted values to certain risks, such as a trial having a new investigator on staff (see chart below). It then runs the different values through a three-part calculation to come up with a risk score. Trials with higher risk scores are more likely to be inspected.

It’ll be interesting to see how, or whether, the automation changes the nature of the agency’s inspections regime, Beltre says.

“How is this different than the old way of selecting sites for inspection? Does the industry need to prepare or do anything different than before? What changes shall we anticipate for sponsors, CROs and sites?” she wonders, adding some answers may not come until there’s more data available.

Linda Sullivan, MCC’s co-founder and president, says shedding light on the process actually reassured trials professionals that the FDA is still focusing on the broader context of trials even though it’s using advanced statistics to inform its inspections.

“There’s been a lot of anxiety around inspections,” Sullivan says. Melinde “calmed a lot of fears. While we may see things in the numbers, we do spend the time to try to understand the context.”